Construction Research Congress 2014 ©ASCE 2014
319
Eliciting Constructability Knowledge for BIM-enabled Automated, Rule-based Constructability Review: A Case Study of Formwork 1
2
3
Li Jiang , Robert M. Leicht , and Gül E. Okudan Kremer
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
1
Graduate Researcher, Dep. of Architectural Engineering, The Pennsylvania State University, University Park, PA 16802; email:
[email protected] 2 Assistant Professor, Dep. of Architectural Engineering, The Pennsylvania State University, University Park, PA 16802; email:
[email protected] 3 Professor, Dep. of Industrial and Manufacturing Engineering, The Pennsylvania State University, University Park, PA 16802; email:
[email protected] ABSTRACT The concept of “constructability” has been raised to optimize construction knowledge and experience in the design phase of construction projects to improve project performance. Previous research demonstrated the feasibility of an automated, rule-based constructability review through BIM implementation. The current work investigates the elicitation of constructability knowledge to achieve the automated process. With a case study, two frequently used knowledge elicitation methods interview and document analysis - were applied to collect constructability knowledge on formwork selection for a concrete building project. Advantages of each technique indicate the need for combining the two elicitation methods as a means of triangulation to establish a comprehensive and accurate knowledge base for an automated, rule-based constructability review. INTRODUCTION The constructability concept has been raised and provided substantial opportunities for optimizing construction knowledge into the design phase. Due to an increased awareness of the potential benefits in improving constructability, many construction companies began to conduct constructability reviews at different design stages to improve the reliability of design and facilitate the construction process. By capturing constructability knowledge from construction experts, a checklist and a lessons-learned system are frequently adopted after the design reaches a certain design stage, 30%, 60%, or 95% design (Hancher and Goodrum 2007). However, the large amount of required resources (i.e., time and manpower) largely impedes constructability implementation (Hancher and Goodrum, 2007); the rework caused by the inefficient process (Pulaski and Horman, 2005) cannot be ignored either. The idea of implementing integrated design methods to enhance project performance, notably leveraging Building Information Modeling (BIM), has been greatly pushed for project design as a shared knowledge resource of a facility among project participants (Rekola et al., 2010). A recent study of a new dormitory building at The Pennsylvania State University concluded that BIM has great potential to perform an automated, rule-based constructability review of a design model (Jiang et al., 2013). Building on previous study, this paper investigates the knowledge elicitation process to develop constructability rules and enable the automated review
Construction Research Congress 2014
Construction Research Congress 2014 ©ASCE 2014
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
process. Based on a case study, two frequently-used elicitation techniques - interview and document analysis - were implemented to collect the constructability knowledge for a concrete building structure. By comparing the two approaches, the means of translating case-based constructability issues into rule checking parameters is presented. The following sections review previous research, upon which the basis for the current work was constructed. KNOWLEDGE CONCEPTS AND ELICITATION Knowledge is what we know, the sum of what has been perceived, discovered, or learned through experience or study (Schubert et al., 1998). Based on a taxonomy of learning outcomes, knowledge can be categorized into four types: (1) factual, (2) conceptual, (3) procedural, and (4) metacognitive. The first two types-factual and conceptual- constitute knowledge of “what;” the latter two types-procedural, and metacognitive constitute knowledge of “how to” (Anderson et al., 2000). Knowing “what” helps to solve convergent problems, for example, determining which type of formwork will be used, given a list of constraints. Knowing “how to” is most useful for designing a plan or sequence of actions to solve a problem, such as how to make formwork decisions step-by-step. Acquisition of knowledge on “what” and “how to” builds the knowledge resources in support of the decision-making for a constructability review. The first step of acquisition is Knowledge Elicitation (KE) (Diaper, 1989). The distinction between the four different types of knowledge implies that various elicitation techniques are needed for soliciting the proper type of knowledge (Diaper, 1989). To capture the strengths and weakness of each technique, criteria used can be related to its operation – brevity, productivity, and accessibility of knowledge source, and its outcome – results validity, breadth of knowledge, and differential access (Schweickert et al., 1987; Hoffman et al., 1995). Regardless, interview and document analysis were found to be most frequently used (Cullen and Bryman, 1988). CONSTRUCTABILITY KNOWLEDGE AND TOOLS Constructability knowledge resides in the heads of construction experts, dispersedly. Hanlon and Sanvido (1995) classified this constructability knowledge into five categories of concepts: (1) design rules, (2) resource constraints, (3) performance, (4) external impacts, and (5) lessons learned. In order to better incorporate the constructability concepts as project design develops, Pulaski and Horman (2005) defined constructability knowledge in different levels of detail, from building/site level to elements level, to gauge what constructability knowledge should be introduced at each design stage. With the help of information technology, previous research has developed knowledge-based tools to facilitate constructability implementation at project design (Fisher, 2007; Jiang et al., 2013). For most tools, interviews have been used as the elicitation method to collect the related constructability knowledge, offering a decision making framework for constructability diagnosis of a given design. However, the lack of comprehensive analysis from building/site level to element level, “proactive” or “preventive” feedback at early design, and the visualization support result in confined applications
Construction Research Congress 2014
320
Construction Research Congress 2014 ©ASCE 2014
321
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
in practice (Jiang et al., 2013). Thus, a more powerful tool is needed to improve the current constructability review process. BIM POTENTIAL FOR AUTOMATED CONSTRUCTABLITY REVIEW Since late 1980’s, Building Information Modeling (BIM) has been adopted and implemented in the AEC industry. As a potential tool towards integrated design and delivery, BIM is expected to improve the efficiency and to enhance the value delivered through project lifecycle (Owen, 2009). Based on the technological features and emergent practices (Taylor and Bernstein, 2009), Jiang et al. (2013) presented an approach of using BIM to achieve an automated, rule-based constructability review. The rule-based approach for automated constructability review of a design model can be divided into 4 stages: (1) rule interpretation, (2) building model preparation, (3) rule execution, and (4) constructability feedback (❶-❹ in Figure 1; based on Eastman et al. 2009). With a case study of a new dormitory building at The Pennsylvania State University, the design-related constructability knowledge was captured with related BIM contents, demonstrating the feasibility of the BIM-enabled approach. The previous investigation was preliminary and introduced the framework of the innovative approach. The current work is a follow-up study, investigating the means of obtaining the required constructability knowledge from construction experts and interpreting them into constructability rule-sets (❶, Figure 1). ❶ Constructability Knowledge
Constructability Rule-sets
Building Information Model
❸ ❷ Rule-based Platform
❹
Constructability Feedback
Figure 1. Framework of BIM-enabled Rule-based Constructability Review CASE STUDY Based on a cast-in-place concrete project, the current work conducts a case study with the two most frequently-used KE techniques: interview and document analysis. The operation and outcomes of both techniques were compared to investigate their efficacy of acquiring desired constructability knowledge for selecting the most appropriate construction methods. Project description Located in Maryland, the case study project is a community center with 5 buildings interconnected via an underground parking facility, including a mosque. The complex has a gross floor area of approximately 316,000 square feet, more than 95% of which is constructed with cast-in-place concrete. One of the 5 buildings has a one-story steel structure, and thus was not considered in this study. Given cultural concerns, the project design has incorporated traditional mosque features such as domes and minarets, resulting in a range of different formwork systems used in the project (Figure 2). The variety of the formwork applied makes this project a good
Construction Research Congress 2014
Construction Research Congress 2014 ©ASCE 2014
322
case to capture the constructability knowledge for selecting formwork systems for a given design. Formwork
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
Horizontal
PERI SKYDECK system
Special Structure (i.e. Dome/Arch)
Vertical
HARSCO decking system
Spanalls
PERI TRIO wall panels
MEGAFORM panels
Styrofoam Balls
Customed wood formwork
Customed Falsework
Customed wood stud system
Flying form
Figure 2. Formwork Used in the Case Study Project Knowledge elicitation Two KE methods, which were recognized to be the most frequently-used, were applied independently in the case study: interview and document analysis. For both methods, two knowledge engineers were involved in the elicitation process. To better compare the two approaches, both processes were modeled in Unified Modeling Language (UML) as class diagrams, which employed the representation of objects with attributes and relationships between to clearly display the elicitation of the knowledge regarding formwork decision-making. Five UML relationships were used to develop the model of each elicitation process: direct association, generalization, aggregation, composition, and dependency (See details in Medina et al., 2013). Based on the information depicted in the UML class diagram, the knowledge segments for formwork selection were then represented in a decision tree, so that the selection rules can be extracted. Associated rule parameters that were formed by identified attributes and associated values can then be linked with available BIM content to perform an automated constructability review of a given structural design. Interview Focused, or semi-structured interview was adopted in this case study to allow the interviewee to offer their ideas and opinions in a given topic (Diaper, 1989). The entire session was constrained within 60 minutes and recorded for transcription analysis. Two knowledge engineers carried out the session together, with one leading the interview and results review while the other performing background tasks such as operating a digital voice recorder and transcription analysis. With a set of predetermined open-ended questions, the expert, who was defined to be the project manager of the concrete subcontractor and has 8 years’ experience of concrete construction, was asked to “think aloud” during: (a) the general process of formwork selection (i.e. knowing “how to”); (b) the formwork systems selected for the current project; and (c) consideration of factors that lead to the selection outcomes (i.e. knowing “what”). The three parts were expected to collect the experts’ knowledge and develop formwork selection rules. This three-part “think aloud” protocol was not conducted in a strictly ordered fashion, but its sections were intermingled.
Construction Research Congress 2014
Construction Research Congress 2014 ©ASCE 2014
323
produces
LEGEND
Formwork Company may consult
documents
develops
Formwork Systems
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
Following the standard UML relationships, a process model was developed to show the elicitation of the constructability knowledge regarding the formwork selection of a concrete building structure through transcription analysis (Figure 3). As illustrated, knowledge was collected regarding: the selection process through the communication among multiple project participants, and the selection of a certain type of formwork system (Figure 3). Correspondingly, three types of information were captured and colored in the model (Figure 3): project documents (white), project participants (light grey), and rule elements (dark grey).
Concrete Subcontractor Formwork Shop Drawings is documented in
Horizontal Formwork
80% Construction Documents
review
Project Mgmt. Estimator Supt. Mgr. has
Engr.
Architectural Design
Structural Design
is developed by
is developed by
Project Participant Rule Element
Structural Engr.
Architect
are reviewed by
Project Documents
Knowledge regarding process Knowledge regarding formwork
from
UML RELATIONSHIPS
Constructability Concern -Concern: Slab load impacts on formwork system -Category: Resource Constraints
is decided by PERI SKYDECK system -ProductName: PERI SKYDECK -VerticalSupport: Standard props includes
Slab -Name: Slab -Depth: 13' -Material: Concrete
is used for
Direct Association relates with a verb Generalization – “is a” relationship Aggregation – “has a” relationship Composition – “being part of” Dependency – “depends on”
Figure 3: Knowledge Elicitation - Interview As shown in Figure 3, the concrete subcontractor who made formwork decision for this project was involved when 80% of construction documents were completed. Through intra- and inter-organizational communication, formwork shop drawings were developed and submitted for review and approval. With an example, Figure 3 also shows the elicitation of knowledge regarding selecting a certain type of formwork. Related information was extracted from interview transcription for rule development. These information bits were linked by words such as “because” and “so” or phrases such as “instead of…use….” They were captured as rule parameters that can be expressed in terms of “IF…THEN…” statements. In this example, part of the dialogue between the expert (EX) and the knowledge engineer (KE) is shown as follows: EX: …but slab in between is actually supported differently…instead of using spanalls, on the other side, we have to use a deck tower, which is a 13” slab… KE: OK. But it was driven by the load of the slab. EX: Yes. In context, the “deck tower” refers to the “SKYDECK system” of PERI formwork. The depth of a slab was identified as the design-related constraint that drove formwork decisions. Therefore, based on the section of interview provided above, a rule can be extracted as: IF the depth of slab is 13”, THEN “PERI SKYDECK system” is more appropriate as the horizontal formwork.
Construction Research Congress 2014
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
Construction Research Congress 2014 ©ASCE 2014
324
By doing this iteratively, the knowledge regarding formwork selection for a given design was collected through interview and transcription analysis. Based on the identified rule parameters, Figure 4 presents the acquired knowledge for horizontal formwork selection. In addition to design parameters such as slab slope and slab depth, resource constraints such as crane, labor, and the layout density were also considered in the decision-making of formwork use. On the other hand, one formwork – EFCO beam-slab system that was applied in a previous project was mentioned by the expert as lessons learned to better explain the selection of the PERI SKYDECK system, due to the flexibility of the semi-structured interview process. Nevertheless, the information that is captured through interview is lacking precise description, constraining the application of rules, for example, the criteria of floor height was described as “high” or “low” without precise definition of “high” or “low” (Figure 4). PERI SKYDECK system Yes Yes
Yes
HARSCO deck system
Slightly Less
Crane Needs ?
Relatively High No
Beam-Slab Floor System?
Yes
Dense Layout?
Labor Intensive?
PERI SKYDECK system EFCO beam-slab system
Relatively Low
HARSCO deck system High Horizontal Formwork Selection
Slab Slope = 0 ?
No
Floor Height
Large Low
Slab Depth
13" 10"
No
PERI SKYDECK system with beefier props PERI SKYDECK system with standard props Spanalls
PERI SKYDECK system
Figure 4: Decision Tree for Horizontal Formwork Selection - Interview Document Analysis Unlike interview, which has direct interaction with a human knowledge source, document analysis was defined as an indirect elicitation method of “reviewing or evaluating documents – both printed and electronic material,…,which required data be examined and interpreted in order to elicit meaning, gain understanding, and develop empirical knowledge” (Bowen, 2009). In this case study, electronic plans and specifications, and available submittals of formwork shop drawings were defined as the documentation for knowledge elicitation. The same two knowledge engineers were involved in the process, with one performing the major analysis tasks and the other evaluating the results, to maintain the reliability and reduce the bias of the study. The model of the elicitation process through document analysis was developed and displayed in Figure 5. Likewise, knowledge regarding both the process and the application of a certain type of formwork were illustrated. The three types of information - project documents, project participants, and rule elements captured through the analysis of documentation were colored consistently for comparison (“Legend,” Figure 5). Following the process, the knowledge for horizontal formwork selection were captured and represented in a decision tree as illustrated in Figure 6.
Construction Research Congress 2014
Construction Research Congress 2014 ©ASCE 2014
Formwork Systems
Downloaded from ascelibrary.org by PENN STATE UNIV on 12/08/14. Copyright ASCE. For personal use only; all rights reserved.
document
Project Documents
Formwork Shop Drawings Submittal
is documented in Horizontal Formwork include
LEGEND
Design Drawings
is stamped by
325
Structural Drawings
are issued to
Architectural Drawings is stamped by
is stamped by
Project Documents Project Participant Rule Element
Concrete Subcontractor
CM Project Engr.
Contractor
Structural Engr.
Architect
from Knowledge regarding process
from
are reviewed by
from
Knowledge regarding formwork
PERI SKYDECK system -Layout: 5 x 7.5 Grid -ProductName: PERI SKYDECK -VerticalSupport -AllowableBuildingDesign
Verticalsupport -MRKFrame: 1 -ShoreTower: Every 1000 sq.ft -MPType: MP480 -MPExtensionLength: (8' 6 1/2", 15'9")
Floor-to-Floor Height is designed for
AllowableBuildingDesign -BuildingName: Parking Garage -Component:Beam/Slab -Depth: =
